Leveraging Ai to Power Agronomy Transformation

•

THE PROBLEM

Agronomists and Product Managers are always trying to build the next best products for the growers. However, they usually face a lot of blockers during the research phase looking for trustworthy computational crop models which can be salvaged to design the best solution for the end users. Insights Engine tries fill this gap by building a platform for computational computational agronomy powered by machine learning models.

THE GOAL

The goal of this project is to develop a unique and purpose-built platform for computational agronomy that leads the ongoing digital farming transformation. The platform can help build trust within Agronomists and Product Managers to build better products for growers.

THE DESIGN PROCESS

As a Senior Product Designer I was responsible for design sprint planning, product discovery, user research and visual design. I followed the design thinking framework leading to a 2 weeks (10 days) design sprint process.

UNDERSTANDING THE USER

User Research Summary

1. Insights Engine can be helpful to internal agro scientists, product experts & engineers, with long term vision to make it available to public domain.

2. Insights Engine can provide a unified platform for all Agronomy expertise for all all internal users.

3. Insights Engine should have outstanding user interface with accurate visualisation for the output prediction of the machine learning models.

4. Insights Engine should make the deployment of the ML model API easy and hassle free.

Empathy Map

Pain Points

1. Users find it difficult to browse through several external websites and internally available products in the brand digital ecosystem. It takes up several hours of work time to research and validate the findings.

2. Users find it difficult to enquire about technical documentation related to model architecture, metrics and deployment for internally available products in the brand digital ecosystem.

3. Users find it difficult to build trust on available products without being able to thoroughly understand the model output and validate the output with existing field data available through the growers.

User Persona

Problem Statement

Steve is a Digital Agriculture Product Manager from Lille, France, who works closely with growers, who needs to understand and and test how specific agronomy models work because she wants to know how it can be used to develop better products for growers.

EXPLORING IDEAS

Value Proposition

1. Have easy access to technical API and system documentation.

2. Have easy access to scientific model algorithm documentation.

3. Able to easily filter out ML models by different categories of crop management.

4. One-click navigation to ML model test playground user interface.

HMW Exercise- Summary

1.How might Insights Engine Hub assist Robin with understanding of complete Onboarding of ML models?

- Modular structure of the ML model documentation.
- Easy navigation to the documentation page.

2. HMW assist Adam identify available ML models for specific region?

- Adding filter or search option for users to explore by country or region.
- Able to search ML models by crop name, type of model or agronomy topic.

3. HMW assist Adam verify accuracy of ML models?

Easy and real time navigation to ML model output with variable inputs.
Easy to understand visual output of the complex data output from the ML models.

Initial Ideas

Goal Statement

Insights Engine Hub will let users get easy access to available ML models which will affect Syngenta Agronomists, Product Experts & Engineers by bringing all ML model information under one single platform and reducing browsing time looking for agronomy information and build trust on the capability of the models. We will measure the effectiveness by tracking the number of deployment requests for the models.

TAKING DECISIONS

Information Architecture

User Flow Journey

PROTOTYPING

Wireframes

Usability Testing Findings

Usability testing is a method of evaluating how easy and efficient it is for users to interact with a product, such as a website, application, or device. The goal of usability testing is to identify any issues that may prevent users from completing tasks effectively and efficiently, and to gather feedback on how to improve the user experience. We ran remote moderated Usability Test with 5 participants for our target User Personas.

Here are the key takeaways:

1. Users can benefit from a straightforward information architecture of the models with more visibility for individual individual ML model CTA button.

2. Users find the advanced input field options too daunting and lengthy. Should be a better way to present the options.

3. Users such as Product Managers definitely benefit from the graph view to understand the technical data put out by the ML complex ML models.

Visual Identity

The Colour Scheme component is based on the brand Design system library. This is a cooperative project that allow the global team of designers to establish coherence across several different products under the brand.